Longhorn can't be installed despite extensions being available...

[o10s]

Hello everyone I need help cause I have been at it for 10 hours straight and can't have it working...

So I have installed talos on a single node bare metal using the following image:
factory.talos.dev/installer/954a8ed7acae15ec71e8df744294f0dacbe6458f005a3da880902d83034c44c8:v1.6.7

This contains the following:

customization:
    systemExtensions:
        officialExtensions:
            - siderolabs/intel-ucode
            - siderolabs/thunderbolt
            - siderolabs/i915-ucode
            - siderolabs/realtek-firmware
            - siderolabs/util-linux-tools
            - siderolabs/intel-ice-firmware
            - siderolabs/iscsi-tools

I have also changed my installer like so:

install:
  disk: /dev/nvme0n1
  image: factory.talos.dev/installer/954a8ed7acae15ec71e8df744294f0dacbe6458f005a3da880902d83034c44c8:v1.6.7
  wipe: true

I have also issued the following command to my node:

talosctl --talosconfig=./talosconfig upgrade --image factory.talos.dev/installer/954a8ed7acae15ec71e8df744294f0dacbe6458f005a3da880902d83034c44c8:v1.6.7 -n 10.0.0.31 --preserve

Now all of this comes from the documentation available here: https://www.talos.dev/v1.6/talos-guides/install/boot-assets/#example-bare-metal-with-image-factory

For longhorn to work I have also allowed the longhorn-system ns to run restricted pods following this doc: https://www.talos.dev/v1.6/kubernetes-guides/configuration/pod-security/

I "Think" I have done everything according to the docs, and have followed what I need to follow... however longhorn always complains... no matter what I do... the error is this one:
time="2024-03-26T14:28:53Z" level=fatal msg="Error starting manager: Failed environment check, please make sure you have iscsiadm/open-iscsi installed on the host: invalid mount namespace /host/proc/7280/ns/mnt: failed to execute: nsenter [--mount=/host/proc/7280/ns/mnt mount], output , stderr nsenter: failed to execute mount: No such file or directory\n: exit status 127" func=app.DaemonCmd.func1 file="daemon.go:83"

I am REALLY lost and I have NO IDEA what to do now... cluster is down... so yeah it is severely making me angry...

Please send some help !
Cheers !

[innobead]

cc @c3y1huang @longhorn/qa

Have you done longhorn disk setup? https://longhorn.io/docs/1.6.0/advanced-resources/os-distro-specific/talos-linux-support/#data-path-mounts

[o10s]

@c3y1huang I have an older cluster running on pure VMs (on a truenas scale instance) where talos runs longhorn perfectly fine I have to boot it up again go check the versions it's using I believe the version right before 1.6.0 (so 1.5.4)

However in my current installation I run it on bare metal everything being on the latest version

I will be able to do the describe after my work day is over and I will update it by end of afternoon

Thank you !

[c3y1huang]

Interesting. Longhorn doesn't support Talos before v1.6.0. In the previous Longhorn versions, Longhorn was dependent on some host binaries that don't come with Talos. How did you manage to install v1.5.4 on Talos without hitting those dependency limitations?

[o10s]

@c3y1huang therefore I must have indeed installed longhorn 1.6.0 then as I said it's an older cluster I must have been wrong I need to double check once I have access to the cluster

[o10s]

@c3y1huang my bad the ci was deploying the old version of the longhorn chart... I'm dumb thank you for pointing me in the right direction !